dataset · Global

weather-forecasts

Hugging Face dataset openthomas/weather-forecasts

Claim this listing

Represent weather-forecasts? Send a claim for review so we can discuss listing updates. Claims do not automatically grant ownership or verification.

Data offered llm-inferred: daily high/low forecast values by station and target date · forecast ensemble spread (high/low per date) · forecast reference sigma and member counts · forecast issue and run timestamps · weather forecast model outputs (GraphCast/GenCast tags)
Coverage llm-inferred: Global (catalogue regions field); preview keys include nyc and aus · Target dates 2026-10-08 to 2026-10-14 seen in the schema preview
Resolution llm-inferred: Daily high and low per target date (inferred from schema)
Formats llm-inferred: JSON
Access docs-verified: Public, ungated; MIT licence

Use cases llm-inferred: Compare ensemble temperature forecasts against observed highs and lows for weather-sensitive site works scheduling · Benchmark AI weather model forecast skill (GraphCast/GenCast-style outputs) for prediction-market or forecasting trials · Build a historical forecast-error check on daily temperature inputs before using them in a drainage or heat-risk assessment
Audience llm-inferred: data scientists · forecasting analysts · hydrologists · civils project planners

Constraints llm-inferred: Dataset viewer failed with DatasetGenerationError (CastError on mismatched column names), so the preview is unavailable · Schema suggests mixed record layouts across files, so loading may need per-file handling · No variable descriptions or units are given in the docs; the variables are only identifiable from column names · Preview shows only city-style station keys (nyc, aus); UK coverage is not evidenced · Not a UK water or civils dataset; fit for UK consultancy work is unverified

Verified 2026-10-15 via HF API: exists; 2070 downloads/30d, 0 likes.

Metadata observed 14/10/2026 · research sweep: manual-2026-10-15 (Hugging Face datasets)